Sharp Tech with Ben Thompson - CHIPS History and Pelosi’s Taiwan Visit

Episode Date: August 4, 2022

A brief history of semiconductors in the US and Taiwan, what the CHIPS act can and can’t fix, and why the legislation is probably a good idea regardless. Then, reactions to Nancy Pelosi's visit to Taiwan.

Hello and welcome to the first ever episode of Sharp Tech with Ben Thompson. I'm your host, Andrew Sharp. Ben, how you doing? I am excited. I mean, we've been working on this concept for a while. There were some beta episodes that will never see the light of day. But to actually be doing it for real is very exciting. Yeah, for the record, the beta episodes were good.

They're just now like five or six months old. We're starting fresh. They're legendary. They're out there. People will not know about them. They won't know what they were about. We'll just constantly refer to them and drive people up the wall. That's going to be the plan going forward.

Exactly. Well, now we have a logo. We have a name. It's all so much more official. So this is our maiden voyage with Sharp Tech. Very exciting stuff. You're back in Taiwan now. And I'm ready to get rolling.

I'd say we should christen the podcast somehow, but not really sure how we would do. My only reference for christening is Caddyshack. China christened it for us by sending a missile over my head. There you go. That's exactly what it was. Well, that's where we're going to be today. In part two, I want to talk about Pelosi's visit to Taiwan and some of the potential impact

that that could have. But with part one, I want to talk about microchips. And we'll start by framing the conversation with an excerpt from CNBC this week. After three years of stops and starts, the U.S. House of Representatives passed a $52 billion package last week, designed to boost semiconductor manufacturing in the United States and improve competitiveness with China. The bill passed the Senate with broad bipartisan support and will now go to President Joe Biden to be signed into law. The bill addresses concerns expressed by both parties that the U.S. needs to bring the production of these vitally important components back home and rely less on Asia-based manufacturers. Supporters, including companies such as Intel and Global Foundries, argued that the bill was necessary since other countries subsidized their semiconductor industry, making it hard for U.S. companies to compete without help. Describing Intel, as a supporter and not like a direct beneficiary is kind of fun there.

To say the least. Yeah. So before we get to the bill itself and what it might mean in this country, I do have sort of a threshold question for you. And it's the type of thing that could lead to like a four hour podcast if we wanted it to. How did we get here? Like why did Taiwan and South Korea and the Netherlands,

become like the global leaders in chips and how did the U.S. let itself fall this far behind? Well, I think the important thing is there's multiple steps in sort of the chip value chain. So when it comes to chip design, the U.S. is still by far the leader. I mean, you have the most well-known examples, Apple making their own chips, but NVIDIA is a U.S. company, AMD is a U.S. company, all of the like intellectual property. about what a chip is and how it does is by and large in the U.S. and also on the intellectual property about how you actually make the chips in the U.S. And the whole process, there's a bunch of sort of companies in the value chain.

You mentioned the Netherlands. The Netherlands company and question here is a company called ASML that makes a tool that goes into manufacturing chips. They're not actually designing the chip. That's like one piece of the whole pie. And this bit of the value chain is important because you see. see this in a lot of industries where at the beginning stuff's like super integrated and one company does everything. So if you want to go use Intel as the example, Intel starts out they, you know, they design the chips, they manufacture the chips, they sell the chips, they do the whole thing,

the whole kit and caboodle. I mean, the reason why Silicon Valley is called Silicon Valley is because they actually manufactured silicon chips there. And the chip industry is the foundation of tech. And the fact it ended up in California is this crazy weird accident of history where a guy named William Shockley, who invented the transistor, his mom was like sick. And so he wanted to build a company around this. And he was from Pal Alto. So he, or no, Mountain, Mountain View, I think. So he goes back to Silicon Valley, bring some people with him from Bell Labs. He used to work at Bell Labs in New Jersey. So like the industry could have been in New Jersey, like, like where the transistor was invented. But he wanted to go back to California. Turned to. Turned.

out he's a terrible person to work with. And so a bunch of his employees left, they're called the detraterous aid or something like that. They founded a company called Fairchild Semiconductor. And Fairchild was like they made like radios and stuff like that. They get into the semiconductor business. It's super successful. They kind of don't know what to do with it.

Like, and they, you know, they have other business concerns. And so those employees leave Fairchild Semiconductor to start a new company called Intel. And so Intel is like sort of this direct. chain from that. And as part of Intel, you have... Wait, wait, wait. Just to anchor me, what year was this? These are like the 1960s. Yeah. Okay. Yeah. So 60s. The 60s and 70s. I mean, the transition was invented the 1950s, I believe. I don't have the exact dates in front of me. But like one sort of interesting factoid is there's a guy named Arthur Rock who raised money for Intel. And that was

kind of the... He wasn't the first one, but that was really what the stem. moment in like the venture capital industry. Because you think about something like chips, you have to create the manufacturing capability. You have to actually design them. Like there's a huge amount of sort of upfront costs. But once you're making them, to make one additional chip is practically free. Like there's a massive amount of upfront sort of fixed costs.

And then you have like the sort of zero cost on the back end. And so if you can get huge volume, you can make a ton of money. Like your upside is almost sort of unlimited. And if you think about that, that's how software works, right? You spend a lot of time up front, a lot of money building it. And then if it's successful, you have like infinite upside capacity on the back end. So venture capital actually got started for chips.

And because the chip sort of fundamental economics are similar to software, it sort of transitioned naturally to software. And a big reason why the whole software industry is centered in Silicon Valley is in part because the venture capitalists are there. And the venture capitalists are there because that's where the chips were. And that's how that. So there's a whole sort of like, you asked like, how did we get here? I might be giving you a little bit more of an expansive answer than you want.

No, no. I really appreciate that background because I, first of all, even something as basic as it's called Silicon Valley because silicone is used in the chips. Like I wouldn't have known that like a year ago. Yeah, we say Silicon, not Silicon. Silicon is used another. Silicon. Silicon. Okay.

There you go. Yeah. That's right. But let me ask you, though. So what I was really asking is, like, why, if these components are so vital to our economy, where along the way did we just, like, fall way behind the rest of the world and specifically Taiwan? Well, if you think about it, one of the downsides of silicon being manufactured in the valley is you now have toxic waste sites there. Like, there's a bunch of them. And, you know, it's a dirty, nasty sort of business to actually make this sort of stuff.

And so what happened was you started to get a lot of the production shifted elsewhere. And Intel was definitely a leader in this regard. And they have fabs in Oregon. They have fabs in Arizona. They have fabs in Israel. They have fabs in Ireland. They have fabs in China. And so they have a huge packaging sort of thing in Malaysia.

And it's, it's all. almost a story of globalization that we've seen in industry after industry, where you have the situation where the super high value ad stuff is the design. It's actually like figure out how it works, very sort of white collar sort of thing. And the actual production is it's manufacturing. And manufacturing has been going abroad for decades. And there's lots of factors that go into that.

one underappreciated thing is currency. Like one of the prices of having the reserve currency in the world is your exports are going to be challenged, relatively speaking. And so it's going to be cheaper to do abroad. There's labor issues like in having, you know, can be cheaper abroad. There's environmental issues like these can be dirty sort of outcomes and plants, particularly, you know, in the earlier years. And so for lots of reasons that are kind of applied to lots of industry,

the manufacturing pieces went abroad while the super high value stuff sort of stayed, you know, stayed in the U.S. So that's a broader picture of why manufacturing shifted. The most important development, though, was the creation of TSMC. And TSMC really is a special and unique and one of the most important companies in tech history. And the reason is because what TSMC did, they came along. This is the 1980s. And Morris Chang, who had worked at Texas Instruments for many years, was brought back Taiwan. They're like, look, we want to create an industry in chips.

And he looked around and he said this quite honestly in interviews. He's like, we had no advantages. Like, we had no capabilities, nothing we could do. But what he saw when he worked at Texas Instruments was there were people that wanted to design chips if you have to create your own manufacturing capacity, the barrier to entry to design chips is absolutely massive because you have to get all that stuff. So Texas Instruments would have spare capacity.

And they're like, oh, you can come build your stuff here. Intel would have spare capacity. But the problem is you're not first priority, right? Like if they have to use up your capacity, oh, sorry, you got kicked off. Also, you're giving them all your IP and maybe they'll borrow it. Maybe they'll take a look at it. there's just lots of reasons why it wasn't ideal and it didn't work well to go with these IDMs,

integrated device manufacturers for manufacturing your chips. And that is where Morris Chang saw the opportunity for TSM. He's like, what we will do is the only thing we will do is manufacture chips. We will make a promise to our customers. We will never compete with them. We will guarantee their capacity. We're not going to come in and kick them out. And we're way behind.

Like our capabilities are multiple generations behind an Intel, but you have these advantages that you can't get elsewhere. And the reason why this was so important and impactful is once TSM existed, it created the conditions for this explosion in new kinds of chips and new kinds of chip companies. Because now the barrier to entry was not creating your own manufacturing capability, which was getting more and more expensive. as time goes on as, you know, to get to be sort of an advanced chip. But also, like, you had to have a guaranteed amount of volume to get, like, people to pay, you know, to make that sort of investment. I mean, you're right. It could be for our podcast.

But a key thing about this manufacturing is you're spending all this money up front. You have to have full utilization of that. Or all's like your, your economics will sideways sort of very, very quickly. And so you would only get investment in a big opportunity. And so if TSM's there, they've de-risked all that. And they've decreased the barrier to entry to do this because you can say, look, we're going to raise money to make a chip startup.

Our only costs are going to be sort of the software we need, you know, to design the chips, the engineers who are going to actually do the designing. And that's it. Because once we design it, we can outsource it to TSMC to build it. They've already taken all, they've de-risked that sort of manufacturing bit completely. it's a similar thing to like an AWS for example. That's yeah, that was going to be the analogy for me.

As you explain all that, I'm hearing Amazon story too, where they made it a lot cheaper for companies to use their hosting service and then scale up their business to the point where then they would have the resources to bring it in-house. But with Amazon basically having like this massive scale

to cover those costs for people, it helped the entire internet grow. And it helped everybody basically use Amazon and sort of centralize that piece of the internet economy. Correct me if I'm wrong on any of that. No, that's exactly right. Amazon, AWS transformed tech. And the reason it did it is for exactly what you said.

Instead of having to raise a whole bunch of money before you built anything to buy a bunch of servers and a bunch of infrastructure so you could make something, you could literally in your bedroom make something and you could put it out there and be available to anyone and you can sort of proof your concept before you raise money. And so you had an explosion in startups because that upfront infrastructure investment, you could now rent it and you could only use what you needed. And Amazon would give you lots of credits like you can use it for free because they knew if you were successful, then they'd sort of like have you, right? Right. And so it's the same exact same dynamic. The big, you know, TSM and AWS, it's basically a perfect analogy.

It's really the same sort of idea where it dramatically decreased sort of the barrier to entry. Now, it's a little bit different because Amazon is buying the same servers, relatively speaking, that you would buy as a startup, right? TSM was pretty crappy in making chips, right? But the fact it just existed was sort of a big deal. this is back in the 80s, 90s, and they sort of just, and they just sort of kept at it. And as you had more things,

like graphics chips is a great example, right? Computers back in the 80s and 90s, all of the graphical processing was handled by the central processing unit, like the Intel chip in there. And so it was called software rendering. It would do the whole thing. The idea of a dedicated graphics card was, you know, like,

Nvidia comes along in the 90s. And they're like, we're going to have this car that is going to be independent of the CPU. It's going to only do graphical processing. Why couldn't Nvidia do that? Because a company like TSM existed. And so, and, you know, similar offerings sort of along the way. I think they actually start with text instruments, but the point holds. And so you make this possibility to create dedicated chips that can do all these specific functions.

And it was okay that they weren't on the fastest process. like an Intel was, because an Intel general purpose chip, because it's a general purpose chip, was slower at doing it. If you had a dedicated chip that was focused on just doing graphics, even if it was built on a slower process, it would be faster than overall. And so you have more and more examples of this.

And what this does is TSMC more and more scale, gives them more and more money. The more money they have, the more they can invest on sort of catching up. They can do these sorts of things. And slowly but surely, they sort of like, they crept up,

crept up, crept up. There's a separate story here about Samsung, who is sort of following a similar path. Samsung came into chipmaking via memory, which is, and actually it could be a whole other podcast, like the way they aggressively invested.

Chips in general are very cyclical industry because it takes a few years to come online, right? And so like the past few years, there's been all this, oh, chip shortage, chip shortage. So we're on the edge of entering a chip glut Because everyone started building all this capacity And then the pandemic is it's going to be kind of a big mess

In the chip industry for the next little bit But what Samsung did in memory His memory was particularly susceptible to this Is every time there was a chip glut And prices dropped, they would double down on their investment Because what they realized was All their competitors would pull back investment Because they're getting killed

Because all the chips price are super low And when the cycle turned, Samson would have that much more capacity that much more ahead. And it's actually a very impressive story sort of in its own right. Anyhow, this might be too much detail. But what happened with TSM? Well, let me just add

one thing. I have a question for anybody who isn't following this that closely or hasn't been over the last few years and doesn't really know the story of TSM. When Morris Chang gets called back and you say they call

him back and say we're interested in making microchips. That's the Taiwanese government, correct? Yeah, so the Taiwanese government played a part in the creation of TSM. Actually, the one external company that did is Phillips. This is another Netherlands connection. Phillips, like, wet go of their shares a long time ago, but they actually have a pretty significant role in the creation of this entire chain.

ASML comes from Phillips. There's a few other examples there. It actually is surprisingly super important company in this. this entire story that no one really sort of thinks about. But, but so it's interesting, though, you mentioned they call Morris Chang back. Because one of the, the critical point for TSM was in the 2000s, Morris Chang retired. I mean, this was like, he started TSMC, I believe in his 50s. Like, like he'd already had a full career sort of in the U.S.

And so he's, you know, he steps down. And 2008 comes along and the financial crisis and the great recession. and TSM announces plans to kind of really scale back their capital expenditure, their building going forward, the sort of thing that companies do in a big recession. And Chang was outraged. So he storms back in. Wow. He takes TSM back over.

And he doubles down on the investment. And the reason was, was in 2007, the iPhone had come out. And what he saw and realized, was one, again, to go back to the Samsung example, if you can stomach it, investing through a downturn is how you take share in sort of this industry. Like it, but it's tough because we're talking billions of dollars, right? Like, it's a high wire act for sure. But number two, the explosion of the smartphone created this massive opportunity to get unbelievable scale,

Intel level scale, in fact, far more than Intel level scale, that who, whoever was going to win that was going to be in a dominant position because you get this feedback loop. If you have huge scale, you're making tons of money. You have tons of money. You can invest in sort of what is next. You win on what's next. You take more share.

You get way more money. And you get this positive sort of this feedback loop. Because again, the way chips work where you're investing so much up front that it's very bifurcated. If you don't get volume, you'll go out of. business and lots of lots of chip companies got on a business. Like there's been a consolidation

and like only a few companies now are at the cutting edge. But if you win, you're going to, you're going to be massively advantaged and sort of the next cycle going forward. And what he realized was this was the greatest of all opportunities

to get TSMC at absolutely massive scale. And that's exactly what happened. TSM really just comes to dominate the sort of smartphone era. And what happened from Intel is they were never in a phone because Intel made a decision very early on that the way we win is because we always have the best manufacturing. The problem with mobile is Intel didn't really care about power consumption that much. Like their chips would use a ton of power and it didn't matter because your computer is plugged into the wall.

Like when that decision was made, it was plugged in. And so Intel did stumble a bit with laptops, but they, again, thanks to their manufacturing advantage and thanks to a new architecture, they did come up with a killer laptop chip. And so they were good in PCs. But phones were a completely different ballgame. Phones, like the iPhone was so small. And that battery was so small and it wanted to do so much that the nothing mattered more than power efficiency.

And an interesting sort of factoid. about the iPhone is the first iPhone chip, Apple actually under-clocked it. It ran about as 70% as fast as it was capable of because they needed power efficiency way more than they needed performance. Yeah. Even though the iPhone was slow. If your phone does not work for an entire day without having to be recharged, you're going to be pissed off and buy a different phone. Yeah, no, exactly. And so you have TSMC sitting there ready to manufacture this stuff. And their manufacturing isn't advanced at Intel, but because the chips are custom made for mobile

and the whole software stack is custom made for mobile, they're still way more efficient and usable in these smartphones. And Intel is, and then smartphones take over the world. Like, there's billions of them sold. TSM. Because you're getting that cycle now.

Now they're getting all the money, all the investment. And so they're catching up, catching up. That was the key sort of flip. Where it really went to, you know, 11 was Apple basically and TSM together. They're like, okay, the future, to get down to the fastest possible chips, we need this new technology called Extreme Ultraviolet lithography.

That's where ASML comes in the Netherlands company. They made this, Intel invented it, by the way. They were developing this capability to use this new kind of patterning. When you're designing a chip, you actually shine light. and you like trace all the things on there. But it has to be like super thin. An extreme ultraviolet like there's, it's crazy how it works.

Like, yeah, explode like a drop of, uh, of like tin with like a laser and then you like the extreme ultraviolet that used them. It's insane. All this shit is why building a foundry costs like $15 billion.

Right. No, exactly. Exactly. The entire governments to invest in all of it. And so what, and so basically it, it,

it didn't work very well. And the yields were low. And Intel's like it's not ready yet. We can push our current technology, which is called immersion-based lithography. We can push it to one more level. And what TSM did was they took the leap. And the reason they took the leap, which was a massive risk and costs tons and tons of money and tons of time. And because the reason it costs so much money is you don't have to buy the machine.

But for the first while, your yields are very low. So you're making 100 chips and only 10 of them work. And like, you have to get up to much, much, much higher yield. the reason they could do it is because Apple was with them. And Apple's like, when you pull this off, number one, we'll give you a bunch of money as you need it. I mean, these are all, you know, unconfirmed. But also, once you succeed, we have tens of millions or hundreds of millions of iPhones that will take these chips. And so they had like a guaranteed buyer.

And so they make this leap. They come out with the fastest chips in the world. And Apple buys them all. for the iPhone. And Intel has yet to produce an EUV-driven chip. They're working on it, but that's where TSMC sort of like then leap past Intel. But the reason they could do it is because they had the money. They had the volume.

They had all the scale. And they had a customer at Apple buying up everything that they could produce in that realm. That makes a lot of sense. It removes any risk from like diving into that pool. Yeah. There's tons of other stuff about here about why Intel went sideways. Oh, don't worry. There's going to be a separate Intel podcast where we understand the various layers of dysfunction

that have defined that company's recent history. And look, they've been massively successful in all sorts of different ways over the last 30 years. But that's sort of a separate conversation to me. There is one point, though, that gets into the Taiwan question. So one of the ways that TSM, why chip fabrication is so difficult is it's not just about buying the machines. It's not just about the money. It's about like there's a learning curve that you have to go down. And there's a sort of collective knowledge that makes this possible where all the different people that work on this, they just figure out, they figure stuff out.

Like you can't leap from 28 nanometers to five nanometers. You have to go from 28 to 14 because you figure out how to make 14. You have to go from 14 to 10 because you've got to make 10. You have to go from 10 to 7, 7 to 5. Like there's just, there's so much, what's the word I'm looking for? Like knowledge that you gain that you can't write down. It's just stuff you know, you know how to do. And this is an organizational thing.

There's no one person that knows all this sort of stuff. And so the way TSMC pulls this off is they are completely centralized in Taiwan. And so all of these people that know how to do this. stuff are here. And so when they build a new chip, they're all around. They can figure out how to get it to work. The way Intel did it was Intel would get it working once, I think, in Oregon. And then they would replicate that foundry down to every single piece of equipment, exactly where it is, exactly that was done in other places in the world. And so that's how Intel could be spread around the world. But one thing that's interesting is it made Intel pretty rigid because that's how, that's

how their sort of model worked, whereas the TSNC model was a little bit more flexible because the rigidity was that it had to all be in Taiwan. And so where the rigidity is applied in the process was different between the two companies. And it turned out particularly when you're making the leap to EUV, the Taiwan model was better. And it's also better for if you have lots of companies coming in with different designs, you need to slightly change things for every different customer, like the whole sort of

foundry mentality, you want to have super flexibility, which means you need, but you have, you need rigidity somewhere in the system. And that rigidity is it's all in Taiwan. And so this gets at some of the geopolitical issues, which is as we shifted to this world where having this dedicated foundry model was the better way, because that's how you could get the scale necessary to make these sort of investments to push forward on things like EUB. The company that had the best structure and approach to do that was TSM. A critical part of what makes TSM work is it's all in one place. All their engineers are all together and they sort of figure this sort of stuff out.

And all that would be fine and well. Like the way the natural state of the world. It would be great. It just so happens that they're situated like, I don't know, a hundred miles from a international superpower that's been fairly hostile recently. I mean, there were fighter jets flying over there last night. That's it's kind of a dicey situation right now. Well, and that's the thing, right?

And this is the big challenge, which is the natural state of affairs is that TSMC is just basically dominant as far as advanced shipping goes. Because you have all these feedback loops that are making them not just better today, but much better place to sort of invest in the future. Over the years. Not just knowledge, but also the money and the market, like all these sort of things sort of compound. And so the question. is, yeah, the geopolitical question, which is China has designs on Taiwan. And it's not recently. Like, they've had designs on Taiwan for as long as the, the CPC has been, you know, the Chinese Communist Party has been in power. And that's like what happens if Taiwan is destroyed. Like the U.S. loses that sort of capability, the West in general, the whole world.

And China does as well, by the way. And so. there's this crazy thing where TSM is in the riskiest place possible. And at the same time, TSM is arguably the single biggest reason why there will not be a war. Because the entire world is just too dependent on this one company that no one could afford it. Certainly for right now, that's exactly right. Because, well, knock on wood, who knows. But at the same time, you look at the damage that China would, incur, they would take a even bigger hit than the United States if TSM goes away.

And that may sort of save us from having to investigate what China, United States. Because one of the one of the reasons why, like, Huawei, for example, was cut off from TSM. And that actually made me very nervous at the time because it's decreasing China's dependence on TSMC. I mean, the U.S. is very dependent on TSMC. The, you know, every smartphone chip is made, you know, Nvidia, Apple, AMD, all these companies have a single point of failure, which is TSM.

Now, Intel is making moves to get in the game, the Foundry game, where you can give them your design and they'll manufacture it. That is hard to do. And it's hard to do for lots of reasons, not just because of the money involved and all those sorts of things, But Intel culturally has always been we do everything, right? And we tell you how it's going to be. And we're going to make the fastest chip and you're going to take it and like it.

And this idea that you're going to be super flexible and work with partners and, you know, let them do make decisions and you're going to accommodate them. It's just very foreign to the way Intel operates, right? Well, no, I appreciate all of that history because what I wanted to have is something of like a resource that we can return to and direct listeners to just so that they sort of have some sort of grounding in all of this because it's going to continue to be one of the biggest issues in the world and certainly in tech. And when you look at the U.S. side of things, first of all, my favorite revelation of this podcast was Morris Chang coming out of

retirement like Jordan after baseball and storming back to say, Listen, we are not going to miss the mobile opportunity. Let's double down here. That's phenomenal. But back in the spring, Morris Chang was interviewed by the Brookings Institute, and he said this. We think that the recent effort of the U.S. to increase onshore manufacturing of semiconductors. Right now, you're talking about spending only tens of billions of dollars. of money, of subsidy.

Well, it's not going to be enough. I think it will be a very expensive exercise in futility. The U.S. will increase onshore manufacturing of semi-tides somewhat, but all of that will be very high cost increase, high unit cost. It would be non-competitive in the world market when you compete with factories like TSM. So do you agree with that assessment?

Are we ultimately looking at sort of a doomed experiment here? And if so, are there other reasons that the Chips Act makes sense to you? Yeah, I do agree. I mean, just to put it in context, the Chips Act is, what, $50 billion or something along those lines? The three nanometer fab that TSM is ramping up for production is costing $20 billion. One fab. And so $50 billion is a drop.

TSM is spending $40 billion on capital expenditures this year. And so $50 billion spread across a number of companies is a drop in the bucket without question. It really is. And moreover, two Changs. point, like TSM tried building a founder in the U.S. previously. It still operates. It's in Oregon or Washington. It's in the Columbia River there. Oregon, yeah.

Yeah. And they, it's one of those things where they keep running it because once you built it, you built it. Like you spent all the money up front. But they never got it to be very profitable, relatively speaking, in part just like the labor costs were higher. The work culture in the U.S. is different than the Taiwanese work culture, I think is one way to put it. Like just lots of reasons why it never was really competitive. And the reason why Intel could get away with manufacturing the U.S. is because that was the only place to get Intel chips.

Like Intel charged a lot. And their chips were really expensive, but they got away with it because they were the best processing chips, right? And so they could absorb higher unit costs along the way. When you're competing, the idea of being a foundry of making chips that other people make is you're giving up the differentiation of it being your own design. Like you're just competing, you're a commodity provider. Like, so do you want to manufacture a TSMC or do you want an extra intel?

Well, TSM comes back and says, well, that's going to cost you $30 a chip or for something cheap. It's going to go, 50 cents a chip, whatever it might be. Intel's going to be like, well, that's going to be $40 a chip. And they're probably making lower margins than TSMC is, in part because it's such a scale play. There is the whole cost issue. There's all these sorts of things going into it.

And this gets to the point I made earlier that all things proceeding naturally, it's folly in many respects to even try to compete with TSM right now. And now, Morris Chang is obviously a biased observer, like to say the least. But that doesn't mean he's wrong in this case. Yeah. Well, and I look at that. It is very important context because a lot of people in D.C., where I am, hear the Chips Act is being passed and, get very raw, raw about just bringing all of this on shore. And when you look closer at all the advantages TSM has and how gigantic the gap really is,

I think everybody should be clear-eyed about what can actually be achieved through all of this. And one point that Intel is right about is these companies do have advantages from their governments. I mean, TSMC actually started with Taiwanese government support, but even today, they have very favorable tax regime in particular, where, their depreciation can be written off or like particularly in buying new equipment. And so that is a big advantage. And I do support that passing the U.S.

There's other stuff like the currency advantage that as long as the U.S. is the reserve currency is always going to be the case. And by the way, Taiwan may manipulate their currency, which definitely helps TSMC. And so there's like there are real advantages that they're getting. But it's key to, note that if TSM at this point had zero advantages, zero tax advantages, zero currency advantages,

zero subsidies, whatever it might be, they would still be in the dominant position because of all the other factors that we're talking about. So it mattered at one point. It doesn't really matter today. Why do it anyway? Well, because of China and Taiwan. Like, that's really the reason to do this, which is China is also investing heavily in their chip industry. their leading, their, their, they're leading, uh, foundry is a company called Smick. Uh, and they're, you know, they're pretty advanced. Like, they've, they've poached a lot of TSMC engineers with very, uh, attractive packages. Like they, they, they, they're producing 14 nanometer at scale.

They had now have a set of a nanometer chip out, out in the market. That's probably about as far as they can get because they're banned from buying ASML's UV machines, which, again, ASML is the only company in the world that's figured this out. and why can the U.S. ban it because it's like a ton of its U.S. IP. And so that's what happens there. But that's still pretty good. And so the worry is what happens if TSNC is taken off the board, right? Like China has SMIC.

They have some sort of capacity there. Smick is, you know, they're going to keep investing. They're going to get better. The U.S. has what? Like right now, we have global foundries. Global foundries stopped at 14 nanometers. Like they gave up on going down because it was just too, expensive. And there's still a huge market in these like slower chips, you know, for things like

cars and like not everything needs to be the most cutting edge sort of processes. But the U.S. didn't doesn't have sort of equivalent capacity here. And so it's, it's, it's, and this is why it's so hard. It's, it's, it's an insurance policy. It's, we need to have sort of our own capacity independent of TSMC. But the real question is, it costs so much money. And if there's no real motivation for a company like Apple or Qualcomm or a AMD to partner with Intel, then like, how do you actually get them up to speed in a way that, you know, that that makes this, you know, viable. Like, dire. Yeah.

If the insurance, if the insurance policy is never asked for, right? If, and so it. Well, let me ask you this, though, as far as where China is right now, as we record this in the first week of August. They can make seven nanometer chips. And by the way, as an outsider who's pretty new to this world, the smic, the full smic name, semiconductor manufacturing international corporation is just incredible to me. It's the most generic company name you can possibly imagine. It's like a bunch of mid-70s Soviet architects decided to name a company. And Smick, they now have this, the capacity to make seven nanometer chips,

but to be clear, China would still be screwed if they couldn't buy the smaller chips from TSM. Is that right? Like, they're not in a position where their capacity is so robust that they wouldn't feel the hit. Like, we still have...

Yeah, it's, it's complicated. It's hard to say. I mean, like, like, while it was definitely, you know, hit hard by being cut out from TSM, the, you know, like it's all relative, right? Like, like, I mean, I think we use sport one, an iPhone six.

Like, like, I mean, at some point, like, do you actually need the most, the most sort of advanced chips for lots of applications? Not, not necessarily. That said, you know, how much capacity SMIC has is an open question. You know, there's lots of other pieces that go. Like, there's a whole value chain here. Like, there's the, like, to actually create a, you need not just like the ASML machines or the older machines also made by, by, by, by Nikon. in canon, but also you need like etching machine, like there's

web research makes this sort of thing. They're like, you need like these, these, you have to need the actual silicon manufacturing itself. Like, there's all these sort of steps that go into it that China does not necessarily have the, the capability of doing today. And

it's going to be very hard, it's very hard to develop. And China actually faces a similar challenge because if you can just buy the stuff from people that are already doing it well, why would you, like, why buy it from the from the new kid on the block, as it were, right? Like, like, Smick is doing well,

but Smick's buying all their equipment from U.S. companies or, or, or, or, or, or, or Japanese companies or whatever it might be. And the actual, like, Chinese companies that can make those other pieces are lacking. And so China has, is definitely, is very exposed on this front. They're investing tons of money. But, again, it's not a product we solve with just money.

Like, yeah, there's a learning curve. Like, China is a bit, like, a weightlift. that only works on their arms, never works on their legs, because, you know, China came up in this free, free market, open trade, free trade sort of environment. And at every sort of choice, they sort of gravitated towards more labor-intensive industries, because that was their sort of their big advantage. And today, labor in China is actually pretty expensive, but there's so much infrastructure around this that it's still the best place to go, right? You can't, there's no other place in the world you can make iPhones at the scale Apple makes them other than China.

But what happened was you had all the other countries in the world. Well, we can't compete with China on these labor industries. So we're going to invest these super capital intensive, IP intensive industries. So we're going to focus on making camera lenses or camera sensors or chips or whatever it might be. And it all worked because China would buy that stuff abroad. All these other multinational companies would send the labor instances to China to get made. and it all sort of worked together. There's a system, but you have this endgame where it's kind of well documented

that Western companies have lost a lot of labor-intensive manufacturing. But China never really developed the super IP capital intensive manufacturing either because they've been able to buy it the whole way. And so in many respects, it's a problem. It's also sort of a good thing. Like, it's kind of like mutually assured economic destruction. Exactly. Both sides are so dependent on each other because for the last 30 years, they've operated as if this arrangement will work forever.

And that more than anything is the biggest reason why, you know, scare mining about there's going to be a war tomorrow is overstated because there would be such devastation on both sides that it's just not worth it. Right. Well, and that's an important point right now because people reacted to the seven nanometer news. earlier this year and Smick's newfound capacity. And there were certain people who overstated what that actually means now. But the other question I have about the Chips Act, theoretically, is it still a success if the provisions in that bill keep some of these companies from expanding in China and prevent them from sending any of this equipment to China?

to me it looks like we're at least recognizing that this is a long-term threat and trying to mobilize. And if we can constrict the growth of China's industry, then ultimately it's a win regardless of what happens to semiconductors onshore. Yeah, I mean, I think, yeah, with any sort of act like this, you're artificially changing the market, right? You're influencing decisions. and that's by and large why traditional, you know, like classical economists are opposed to any sort of government intervention because it's like you're not getting market outcomes, you're going to have dead weight loss. It's not going to be as efficient as it might have been otherwise.

There's an aspect here where that's kind of the point, right? Like the natural outcome is for more and more production moving to China for lots of these sorts of things. That's a big market for chips. You know, also, you know, there's still some of the labor issues. There's all those sorts of things why it's advantageous to do that. And you want to make that more. expensive and and incentivize companies not necessarily make those choices. And you want more

production of particularly of these sort of advanced foundries to be somewhere other than Taiwan. And being in the U.S. is a place to do it. Now, I, the Chips Act will apply to investments that say TSM or Samsung want to make in the U.S. And that's a good thing because they're, they're good at what they do. The biggest beneficiary is going to be Intel. And it's kind of have a tough thing because Intel in many regards is in bad shape these days. And it really came out in their earnings, you know, that just came out last week where, you know, it was funny. I've been writing about Intel being in trouble literally since the first month of

Shrekory. One of my first articles was about Intel needing to switch to foundry. And meanwhile, for the last decade, Intel stock's been up. Their profits have been great. And people are like, like, what do you, like they're, they seem to be doing fine. The problem is this industry, it's like the investment necessary and the choice is necessary are years and years and years ahead of time of sort of sort of where you're going. And the problem is that Intel has been making bad choices and has not been adjusting

and has been locked into a particular way of doing things for way, way, way too long. But there's no real candidate other than them to sort of be an alternative to CSMC, to sort of like, like the only companies pursuing the cutting edge right now are Samsung, TSM, and Intel. Those are the like global foundries gave up. IBM gave up, you know, far, far more global boundaries did. Obviously, Smic is pursuing it, but they're restricted from the sort of what they can get to do so. And you're not going to start a new company, a new foundry from scratch. Like, like, it's just, we're too far along.

Like, like, it's, it's too late for sort of that, that approach. And so Intel is the best possible conduit for U.S. having this competitive manufacturing capability to match our superior design and IP and all these other sort of things. And the problem is Intel's a mess. And so, like, I, it's tough. Like, Pac-Hal Senior took over a year and a half ago. He finally made the call to get Intel into Foundry to making other people's chips, which was, what my article was about back in 2013 when I started trajectory, that they should go in this direction.

I'm concerned, at some point Intel needs to bite the bullet and be like, look, like, our stock is going to be crap. And we just have to make this investment to go in this direction. And they just, they can't seem to do that. They won't give up their dividend. They, like, they're, they're really trying to thread this super small needle.

And it's, it's borderline impossible. for a public company to pivot that dramatically. Is that right? And like that's a major obstacle for them. It is. And, you know, being a foundry is a very different margin profile. Like you don't have that differentiator that of having your own design anymore. And as long as Intel's behind TSM, the way they will succeed is by discounting, by being

cheaper. And like, Intel's always succeeded by being the most expensive by having the biggest margins. And like their success is going to have to be. something very different than that. It may be in the future they can surpass TSM and regain those margins, but that's many years down the road.

The problem is, if not Intel, then who? Right? Again, we're stuck in the natural state of the world is for TSMC to just dominate the space, but to the extent that's intolerable from a national security perspective, you need to go somewhere else.

Let's give it a shot. Well, I'm somewhat pessimistic. I'm kind of pessimistic that it's going to work. Yeah. I think that's a, that is a rational place to be. This is the right thing to do. And it's a long shot that it's actually going to pay off. But when you look at our national interests in a number of different areas, not just

our economy, national security and everything else, like it makes sense to try to make this work. And, you know, we'll see if Intel can pull it off. I'm not bullish after the last 45 minutes of talking to you, but it's worth a try. So I want to shift gears, though, for the final 10 minutes here because Nancy Pelosi visited Taiwan this past week. You landed the same day that Pelosi did in Taiwan. I appreciate all the welcome to Taiwan manners, yes.

It's great. Well, and look, most part two's of these shows are going to be completely different than the topic in part one. This is ultimately just like a continuation of the topic in part one because Pelosi's visit became the biggest story in the world this past week. I want to get your read on the situation on the ground. Can I just tell you as an American watching all of this, what I found most unsettling was the idea that Pelosi could just unilaterally decide to thrust the entire.

country into this precarious maybe standoff with China. She threw a lifeline to President Xi, who can now distract the Chinese people from all sorts of dysfunction in China and point to this common enemy of the U.S. And she totally defied the will of the White House and like the State Department, all of whom we're saying, do not do this, not necessarily publicly, but after the fact, it's been pretty clear that like the White House did not want this to happen. So as an American, first and foremost, I think we need a better process of like checks and balances. No, as an American, you should be celebrating this. The reason she could go is because we have separation of powers and there's jack all that the president can do to stop her.

Yeah, well, I mean, I mean, I get that it's a double-edged sword, but I would also like to get everyone on the same page before we decide to provoke an international superpower. Well, a couple things. It's worth pointing out that China can decide. China has agency here. Like, there's so many discussions are like, look what we did to China. And China's like, look what you made me do. Like, they're the living embodiment of a Taylor Swift song. Like, and the truth, and so many people in the West buy into this.

And it's this very American narcissistic view where everything in the world is a function of what we do. And it's like, no, China has choices here. They could have. Okay. ignored it. They could have pretended nothing's going on. Congressional delegations come to Taiwan all the time. And like to manufacture it in this huge thing, the third person in line for the presidency,

it's like, well, you got that bit about the Constitution. Did you get the part about separation of powers? Where like, like, is Pelosi doing it for domestic political gain almost certainly? Like, like China decided to be upset about this. It's worth sort of keeping that, keeping that in mind. Number one. Number two, there is a tendency. just as if I can sort of get on my soapbox here,

to look at China and say, you know, China doesn't have this problem. You know, if someone from China wants to go to the U.S., and Xi Jinping says, no, they're not going anywhere. And look at China, look at how they get, they build all this infrastructure. They do that sort of Steve. Boy, wouldn't it be great if we had that sort of like

ability to make people to do what we want to do? You see this with stuff on the Internet, right? It's like, what if we could actually like remove all the bad stuff from the internet? Get rid of all the misinformation, you know, like, then everything would, be great. And it's a false promise one, because that that's never going to happen. Number two, it's this, this assumption that we know it all and that if only we could get our way and tell the stupid people what to do, everything would be fine. But number three, one of the things fueling China's

rise without question is they're catching up. And I don't say this disparagingly. The U.S., what fueled the U.S.'s rise 100 years ago, we were catching up to Great Britain. And you have like the IP debates. We stole all kinds of IP from the UK, right? Like that we built our entire, you know, industry. I mean, like we stole the entire rule of law from the British system. Yeah, exactly, right? So they still cite the common law, right? Yeah. So I don't begrudge China that at all. But there is the question of once you're on the cutting edge, how do you actually drive growth? How do you actually innovate. How do you actually create new things? And part of the brilliance of the U.S. system as a whole, in which I think a lot of us should be a lot more aware of and vocal of and

defend is that all the messiness is integral to and inseparable from new ideas, innovation, new creation, new things. And so I'm going to push back on you a little bit, which is sure, it'd be great if we get Washington to see all on the same page. Two, it's worth wondering whose same page is that, right? You know, does the Biden administration have foreign policy completely figured out? Like, maybe it's useful to have different ways of looking at it, right? And this applies to all sorts of things. And so, yes, it's messy what went down this week.

If I were in charge of the world, I would probably not have done this trip. It's just not worth the trouble. But it's also worth keeping in mind that all the factors that go into this messiness are weaknesses from a certain perspective, but they're fundamental essential strengths from another perspective. And it's good to keep that in mind every time you get frustrated at American messiness.

Yeah, well, and I wouldn't say I'm that frustrated at the messiness. It's more just like there should be some recognition that the potential benefits of this visit are drastically outweighed by the potential costs and particularly at this moment with the, U.S. sort of staging a proxy war in Ukraine, and as that continues month after month and sort of ravages the world economy, it's not the right time to promote China. I'm off my soapbox. Yes, everything

you're saying is correct. Okay. So what's the mood been in Taiwan? Because on the one hand, I'm sure the U.S. support is reassuring. On the other, it feels like the past week or two have sort of escalated tensions and took the dynamics to a more tenuous place, maybe just for the short term. I'm not really sure. But what do you sense when you're on the ground there? Well, I'm a little hesitant to say anything authoritatively. In part one, I'm not Taiwanese. Two, I've not been in Taiwan for the last few months.

You know, I go back to the U.S. every summer. And three, I'm in quarantine. So I'm not talking to anybody right now. Okay, fair. But it is interesting because I think there's this general dynamic where I've been aware of the China a threat for 19 years since I first came to Taiwan. People in Taiwan have been aware with their entire lives. It's an omnipresent sort of thing.

It's always there. And there's a lot of people in the U.S. in particular who didn't even thought Taiwan was Thailand until a year ago. And they've gone from zero to 100 about the risk factor. And all it's going to be Hong Kong all over again, leaving aside the fact Hong Kong was already part of China is contiguously, contiguous with China's landmass. You know, like there's lots of differences when it comes.

of Taiwan and have overshot it, right? And so I'm sitting here, I'm like, look, I've been concerned about China for two decades, way more than you were. Now, you're telling me I'm insufficiently concerned, it's like, go, chill out there. And I think that that's a sense among Taiwanese in general. At the same time, I do get the sense that there has been a little bit of, this has been a little bit of a wake-up call. and Taiwan has not invested the way they need to, I think, in the correct sort of military readiness that goes into responding to Japan.

They want to buy planes and tanks and stuff like that when they need to be buying missiles and mines and things along those lines. I think there's like Taiwan's energy situation is a mess. They're closing on their nuclear. They're relying on imported natural gas. Like now you see this practicing for a blockade. It's like, whoa, that might be a problem. And I think just in general, I'm actually somewhat optimistic that this is a good wake-up call for Taiwan in particular. Like the threat is real.

And yes, you've been saying that. You've known it's real. But have your actions actually reflected the reality of the situation? I would argue they haven't. And so to the extent people are like, look, we really have to take this seriously. I think that this may end up being, may end up being a good thing for Taiwan, even if having missiles flying overhead and, you know, Chinese ships outside the ports is not, not the ideal place to be. Yeah. Well, and I could see how keeping the threat out of sight, out of mind and being in not necessarily denial, but just not really letting it anchor your decision making for 20 years would make sense. but China is just in such a weird place that in the same way that a weakened animal becomes even more dangerous,

it sort of seems like that might be happening here. But for now, and then you have Xi who's, you know, centralizing power, I think is a brilliant internal political operator, but lots of his other decisions don't seem to be very great. Yeah. And so the fact that he's still committed to zero. COVID two and a half years later, maybe it's unfair, but it like makes me question his relationship

to reality in a way that's unsettling as I, as I envision like what he wants for China and where he might take China. And then just sort of go back to my soapbox rant, the complete inability of anyone to challenge that or push back on it. Like the, you end up with some sort of, you know, feedback. loop. And the question is how long is that feedback loop? Ideally,

and ideally, you know, sort of the ideal democratic outcome is the feedback loop is the next election. When you have sort of an authoritarian who doesn't broke any sort of dissent and there's heavy incentives to sort of please him because that's how you sort of

move forward. And so you're going to overdo things and overreact to things. The feedback loop is usually some sort of unmitigated disaster. And an unmitigated disaster is exactly what any sort of conflict about Taiwan would be. And it's she first and foremost, that is the biggest reason to worry that it could go sideways. Yep. Well, time will tell.

Future Sharp Tech episodes are not going to have quite as much intense geopolitical speculation. Hopefully we'll keep things light on our next show, but I enjoyed talking through all this with you. I'm glad you're safe and sound in Taiwan, and I look forward to coming back next time. Sounds good. I'll talk to you soon.